Problem 649

Summary: Processes for electrons are very sensitive to process ordering
Product: Geant4 Reporter: rikard.sandstrom
Component: processes/electromagneticAssignee: Michel.Maire
Status: RESOLVED INVALID    
Severity: normal    
Priority: P2    
Version: 5.2   
Hardware: PC   
OS: Linux   

Description rikard.sandstrom 2004-07-15 12:32:06 CEST 
Hello,

this is a problem that has been reoccurring in two different simulations. The
first simulation is essentially novice example N02, but with different geometry.
The second example is a bit more complex. However nomatter if using the physics
list for N02 or not the problem persists. If this is a user error please tell me
what I do wrong.

The problem:
When shooting electrons of energies between 2.775 to 11.1 MeV onto 30 cm of
liquid hydrogen (which is encased by 0.14 mm Al walls) I notice very different
results depending on how the processes are ordered. For example:

pmanager->SetProcessOrdering(theeminusMultipleScattering, idxAlongStep,1);
pmanager->SetProcessOrdering(theeminusIonisation, idxAlongStep, 2);
pmanager->SetProcessOrdering(theeminusBremsstrahlung, idxPostStep, 3);
pmanager->SetProcessOrdering(theeminusMultipleScattering, idxPostStep, 1);
pmanager->SetProcessOrdering(theeminusIonisation, idxPostStep, 2);
pmanager->SetProcessOrdering(theeminusBremsstrahlung, idxPostStep, 3);

gives for for 5000 incoming e- the following number of processes occuring inside
the liquid hydrogen

e-, eIoni = 5095
e-, eBrem = 492
gamma, phot = 55
gamma, compt = 207

Instead using

pmanager->SetProcessOrdering(theeminusMultipleScattering, idxAlongStep,2);
pmanager->SetProcessOrdering(theeminusIonisation, idxAlongStep, 3);
pmanager->SetProcessOrdering(theeminusBremsstrahlung, idxAlongStep, 1);
pmanager->SetProcessOrdering(theeminusMultipleScattering, idxPostStep, 2);
pmanager->SetProcessOrdering(theeminusIonisation, idxPostStep, 3);
pmanager->SetProcessOrdering(theeminusBremsstrahlung, idxPostStep, 1);

and leaving the rest of the program unchanged the same results are

e-, eIoni = 2097
e-, eBrem = 244
gamma, phot = 32
gamma, compt = 102

I suppose you see the problem.

Cheers
Comment 1 rikard.sandstrom 2004-07-15 12:38:59 CEST 
The maximum allowed step length inside both the liquid hydrogen and the
confining windows was set to 0.5 mm.

The simulations use magnetic fields around 4T parallell with the beam line.
Comment 2 Michel.Maire 2004-07-19 05:09:59 CEST 
MSC must be the first of the processes which have AlongStep action, as it
perform the conversion 'geometrical' to 'true' path length which is used for the
computation of energy loss.
See PhysicsReferenceManual, item MultipleScattering.
Comment 3 rikard.sandstrom 2004-08-19 05:36:59 CEST 
Hello,

thank you for your reply. The solution you give is not relevant to the
problem however, and I thus request this bug to be reopened until a satisfactory
solution is given.

To demonstrate this I am giving you the address to the files which you can
substitute against already existing files for exampleN02 for
Geant4.5.2p02. /afs/cern.ch/user/s/sandstro/PUBLIC/DemoOfPhysProcesses
(This is actually the whole setup, just run "run1.mac".)

As you will notice, the number of bremsstrahlung events and ionization
events changes place if you change the process ordering for the two
processes. During all time multiple scattering has ordering -1,1,1.

A clue to this might be that this ambiguity disappears when you do not
use a user defined maximum step length. However, in all cases I get
different results. Is this a real bug or am I still missing something?

 1e4 e-, 0.5 mm max step, eBrem at alongstep=3:
* Row:  1=e-,  2=gamma,
* Col:  1=UserLimit,  2=eBrem,  3=eIoni,  4=compt,  5=phot,  6=Transportation,
 0         1         2         3         4         5         6
 1 4.356e+06   1.4e+05      3891         0         0         1
 2 1.144e+08         0         0 1.216e+05      2297      1089

1e4 e-, 0.5 mm max step, shifted ordering of eIoni and eBrem:
* Row:  1=e-,  2=gamma,  3=e+,
* Col:  1=UserLimit,  2=eIoni,  3=eBrem,  4=compt,  5=phot,  6=Transportation,
7=conv,  8=annihil,
 0         1         2         3         4         5         6         7         8
 1 4.358e+06 1.403e+05      3304         0         0         0         0         0
 2 1.137e+08         0         0 1.214e+05      2259      1043         1         0
 3       219         1         0         0         0         0         0         1

 1e4 e-, no cuts, eBrem at alongstep=3:
* Row:  1=e-,  2=gamma,
* Col:  1=eBrem,  2=eIoni,  3=compt,  4=phot,  5=Transportation,
 0         1         2         3         4         5
 1 2.359e+05      4055         0         0         0
 2         0         0  1.22e+05      2392      1051

Please see if you can verify this, and figure out why it occurs.
Cheers